Datasets:

liuganghuggingface
/

moldata-toximol

id string	task string	smiles string	SMILES_original string	SMILES_canonical string	structure_key string	structure_status string	source_dataset string	source_table string	source_row_first int64	observation_count int64
4	ames	O=C(Nc1ccccc1)c1csc([N+](=O)[O-])c1	O=C(Nc1ccccc1)c1csc([N+](=O)[O-])c1	O=C(Nc1ccccc1)c1csc([N+](=O)[O-])c1	["O=C(Nc1ccccc1)c1csc([N+](=O)[O-])c1"]	valid	toximol_ames	toximol_ames/inputs.csv	4	1
7	ames	COC(=O)/C=C/c1ccc([N+](=O)[O-])o1	COC(=O)/C=C/c1ccc([N+](=O)[O-])o1	COC(=O)/C=C/c1ccc([N+](=O)[O-])o1	["COC(=O)/C=C/c1ccc([N+](=O)[O-])o1"]	valid	toximol_ames	toximol_ames/inputs.csv	7	1
10	ames	CCCCN(CCCC)N=O	CCCCN(CCCC)N=O	CCCCN(CCCC)N=O	["CCCCN(CCCC)N=O"]	valid	toximol_ames	toximol_ames/inputs.csv	10	1
15	ames	CN(C)CCCNc1c2ccccc2nc2cccc([N+](=O)[O-])c12	CN(C)CCCNc1c2ccccc2nc2cccc([N+](=O)[O-])c12	CN(C)CCCNc1c2ccccc2nc2cccc([N+](=O)[O-])c12	["CN(C)CCCNc1c2ccccc2nc2cccc([N+](=O)[O-])c12"]	valid	toximol_ames	toximol_ames/inputs.csv	15	1
16	ames	COc1cccc2c1C(=O)c1c(O)c3c(c(O)c1C2=O)CC(O)(C(C)=O)CC3OC1CC(N)C(O)C(C)O1	COc1cccc2c1C(=O)c1c(O)c3c(c(O)c1C2=O)CC(O)(C(C)=O)CC3OC1CC(N)C(O)C(C)O1	COc1cccc2c1C(=O)c1c(O)c3c(c(O)c1C2=O)CC(O)(C(C)=O)CC3OC1CC(N)C(O)C(C)O1	["COc1cccc2c1C(=O)c1c(O)c3c(c(O)c1C2=O)CC(O)(C(C)=O)CC3OC1CC(N)C(O)C(C)O1"]	valid	toximol_ames	toximol_ames/inputs.csv	16	1
18	ames	COc1ccc(O)c2c(=O)c3c(OC)cc4c(c3oc12)C1C=COC1O4	COc1ccc(O)c2c(=O)c3c(OC)cc4c(c3oc12)C1C=COC1O4	COc1ccc(O)c2c(=O)c3c(OC)cc4c(c3oc12)C1C=COC1O4	["COc1ccc(O)c2c(=O)c3c(OC)cc4c(c3oc12)C1C=COC1O4"]	valid	toximol_ames	toximol_ames/inputs.csv	18	1
27	ames	CC(=O)OCc1c2ccccc2c(C)c2c1ccc1ccccc12	CC(=O)OCc1c2ccccc2c(C)c2c1ccc1ccccc12	CC(=O)OCc1c2ccccc2c(C)c2c1ccc1ccccc12	["CC(=O)OCc1c2ccccc2c(C)c2c1ccc1ccccc12"]	valid	toximol_ames	toximol_ames/inputs.csv	27	1
32	ames	O=CNc1nc(-c2ccc([N+](=O)[O-])o2)cs1	O=CNc1nc(-c2ccc([N+](=O)[O-])o2)cs1	O=CNc1nc(-c2ccc([N+](=O)[O-])o2)cs1	["O=CNc1nc(-c2ccc([N+](=O)[O-])o2)cs1"]	valid	toximol_ames	toximol_ames/inputs.csv	32	1
35	ames	Cc1c([N+](=O)[O-])cc([N+](=O)[O-])cc1[N+](=O)[O-]	Cc1c([N+](=O)[O-])cc([N+](=O)[O-])cc1[N+](=O)[O-]	Cc1c([N+](=O)[O-])cc([N+](=O)[O-])cc1[N+](=O)[O-]	["Cc1c([N+](=O)[O-])cc([N+](=O)[O-])cc1[N+](=O)[O-]"]	valid	toximol_ames	toximol_ames/inputs.csv	35	1
46	ames	c1ccc2c(c1)cc1c3c2ccc2cccc(c23)C2OC12	c1ccc2c(c1)cc1c3c2ccc2cccc(c23)C2OC12	c1ccc2c(c1)cc1c3c2ccc2cccc(c23)C2OC12	["c1ccc2c(c1)cc1c3c2ccc2cccc(c23)C2OC12"]	valid	toximol_ames	toximol_ames/inputs.csv	46	1
49	ames	Cc1cnc2cc(C)c3c(nc(N)n3C)c2n1	Cc1cnc2cc(C)c3c(nc(N)n3C)c2n1	Cc1cnc2cc(C)c3c(nc(N)n3C)c2n1	["Cc1cnc2cc(C)c3c(nc(N)n3C)c2n1"]	valid	toximol_ames	toximol_ames/inputs.csv	49	1
53	ames	O=[N+]([O-])c1ccc2c(ccc3cc([N+](=O)[O-])ccc32)c1	O=[N+]([O-])c1ccc2c(ccc3cc([N+](=O)[O-])ccc32)c1	O=[N+]([O-])c1ccc2c(ccc3cc([N+](=O)[O-])ccc32)c1	["O=[N+]([O-])c1ccc2c(ccc3cc([N+](=O)[O-])ccc32)c1"]	valid	toximol_ames	toximol_ames/inputs.csv	53	1
57	ames	O=c1c2[nH]c3ccccc3c2nnn1N=Cc1ccc(O)cc1	O=c1c2[nH]c3ccccc3c2nnn1N=Cc1ccc(O)cc1	O=c1c2[nH]c3ccccc3c2nnn1N=Cc1ccc(O)cc1	["O=c1c2[nH]c3ccccc3c2nnn1N=Cc1ccc(O)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	57	1
2	ames	O=[N+]([O-])c1ccc2ccc3cccc4ccc1c2c34	O=[N+]([O-])c1ccc2ccc3cccc4ccc1c2c34	O=[N+]([O-])c1ccc2ccc3cccc4ccc1c2c34	["O=[N+]([O-])c1ccc2ccc3cccc4ccc1c2c34"]	valid	toximol_ames	toximol_ames/inputs.csv	2	1
8	ames	O=[N+]([O-])c1ccc2ccc3c([N+](=O)[O-])cccc3c2c1	O=[N+]([O-])c1ccc2ccc3c([N+](=O)[O-])cccc3c2c1	O=[N+]([O-])c1ccc2ccc3c([N+](=O)[O-])cccc3c2c1	["O=[N+]([O-])c1ccc2ccc3c([N+](=O)[O-])cccc3c2c1"]	valid	toximol_ames	toximol_ames/inputs.csv	8	1
11	ames	OC1c2ccc3c(ccc4ccccc43)c2C2OC2C1O	OC1c2ccc3c(ccc4ccccc43)c2C2OC2C1O	OC1c2ccc3c(ccc4ccccc43)c2C2OC2C1O	["OC1c2ccc3c(ccc4ccccc43)c2C2OC2C1O"]	valid	toximol_ames	toximol_ames/inputs.csv	11	1
14	ames	COc1cc2c(c3oc(=O)c4c(c13)CCC4=O)C1C=COC1O2	COc1cc2c(c3oc(=O)c4c(c13)CCC4=O)C1C=COC1O2	COc1cc2c(c3oc(=O)c4c(c13)CCC4=O)C1C=COC1O2	["COc1cc2c(c3oc(=O)c4c(c13)CCC4=O)C1C=COC1O2"]	valid	toximol_ames	toximol_ames/inputs.csv	14	1
20	ames	Cc1ccc2c(ccc3ccccc32)c1	Cc1ccc2c(ccc3ccccc32)c1	Cc1ccc2c(ccc3ccccc32)c1	["Cc1ccc2c(ccc3ccccc32)c1"]	valid	toximol_ames	toximol_ames/inputs.csv	20	1
26	ames	CS(=O)(=O)Nc1ccc(Nc2c3ccccc3nc3cc(N=[N+]=[N-])ccc23)cc1	CS(=O)(=O)Nc1ccc(Nc2c3ccccc3nc3cc(N=[N+]=[N-])ccc23)cc1	CS(=O)(=O)Nc1ccc(Nc2c3ccccc3nc3cc(N=[N+]=[N-])ccc23)cc1	["CS(=O)(=O)Nc1ccc(Nc2c3ccccc3nc3cc(N=[N+]=[N-])ccc23)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	26	1
29	ames	Cc1ccc(OCC2CO2)cc1	Cc1ccc(OCC2CO2)cc1	Cc1ccc(OCC2CO2)cc1	["Cc1ccc(OCC2CO2)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	29	1
30	ames	Cc1ccc(CCC2CO2)cc1	Cc1ccc(CCC2CO2)cc1	Cc1ccc(CCC2CO2)cc1	["Cc1ccc(CCC2CO2)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	30	1
34	ames	Cn1c(N)nc2c3cccnc3ccc21	Cn1c(N)nc2c3cccnc3ccc21	Cn1c(N)nc2c3cccnc3ccc21	["Cn1c(N)nc2c3cccnc3ccc21"]	valid	toximol_ames	toximol_ames/inputs.csv	34	1
36	ames	c1ccc(CN2C3c4ccccc4-c4ccccc4C32)cc1	c1ccc(CN2C3c4ccccc4-c4ccccc4C32)cc1	c1ccc(CN2C3c4ccccc4-c4ccccc4C32)cc1	["c1ccc(CN2C3c4ccccc4-c4ccccc4C32)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	36	1
39	ames	O=[N+]([O-])c1cccc([N+](=O)[O-])c1	O=[N+]([O-])c1cccc([N+](=O)[O-])c1	O=[N+]([O-])c1cccc([N+](=O)[O-])c1	["O=[N+]([O-])c1cccc([N+](=O)[O-])c1"]	valid	toximol_ames	toximol_ames/inputs.csv	39	1
42	ames	CC(O)CN(CC(C)O)N=O	CC(O)CN(CC(C)O)N=O	CC(O)CN(CC(C)O)N=O	["CC(O)CN(CC(C)O)N=O"]	valid	toximol_ames	toximol_ames/inputs.csv	42	1
44	ames	c1ccc2c(c1)ccc1cccnc12	c1ccc2c(c1)ccc1cccnc12	c1ccc2c(c1)ccc1cccnc12	["c1ccc2c(c1)ccc1cccnc12"]	valid	toximol_ames	toximol_ames/inputs.csv	44	1
48	ames	Cc1c2ccccc2nc2c1ccc1ccccc12	Cc1c2ccccc2nc2c1ccc1ccccc12	Cc1c2ccccc2nc2c1ccc1ccccc12	["Cc1c2ccccc2nc2c1ccc1ccccc12"]	valid	toximol_ames	toximol_ames/inputs.csv	48	1
50	ames	CC1c2ccccc2-c2ccccc21	CC1c2ccccc2-c2ccccc21	CC1c2ccccc2-c2ccccc21	["CC1c2ccccc2-c2ccccc21"]	valid	toximol_ames	toximol_ames/inputs.csv	50	1
52	ames	Cc1cccc([N+](=O)[O-])c1C	Cc1cccc([N+](=O)[O-])c1C	Cc1cccc([N+](=O)[O-])c1C	["Cc1cccc([N+](=O)[O-])c1C"]	valid	toximol_ames	toximol_ames/inputs.csv	52	1
54	ames	OC1C=Cc2c(cc3ccc4cccc5ccc2c3c45)C1O	OC1C=Cc2c(cc3ccc4cccc5ccc2c3c45)C1O	OC1C=Cc2c(cc3ccc4cccc5ccc2c3c45)C1O	["OC1C=Cc2c(cc3ccc4cccc5ccc2c3c45)C1O"]	valid	toximol_ames	toximol_ames/inputs.csv	54	1
60	ames	Oc1c2nc3ccccc3c-2nnn1/N=C/c1ccc(Cl)cc1	Oc1c2nc3ccccc3c-2nnn1/N=C/c1ccc(Cl)cc1	Oc1c2nc3ccccc3c-2nnn1/N=C/c1ccc(Cl)cc1	["Oc1c2nc3ccccc3c-2nnn1/N=C/c1ccc(Cl)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	60	1
1	ames	c1ccc2c(c1)cc1ccc3cccc4ccc2c1c34	c1ccc2c(c1)cc1ccc3cccc4ccc2c1c34	c1ccc2c(c1)cc1ccc3cccc4ccc2c1c34	["c1ccc2c(c1)cc1ccc3cccc4ccc2c1c34"]	valid	toximol_ames	toximol_ames/inputs.csv	1	1
5	ames	c1ccc2c(c1)ccc1c2ccc2c3ccccc3ccc21	c1ccc2c(c1)ccc1c2ccc2c3ccccc3ccc21	c1ccc2c(c1)ccc1c2ccc2c3ccccc3ccc21	["c1ccc2c(c1)ccc1c2ccc2c3ccccc3ccc21"]	valid	toximol_ames	toximol_ames/inputs.csv	5	1
6	ames	CC(=O)ON(OCc1ccccc1)C(=O)c1ccccc1	CC(=O)ON(OCc1ccccc1)C(=O)c1ccccc1	CC(=O)ON(OCc1ccccc1)C(=O)c1ccccc1	["CC(=O)ON(OCc1ccccc1)C(=O)c1ccccc1"]	valid	toximol_ames	toximol_ames/inputs.csv	6	1
19	ames	Cc1cc2c(c3ccc4ccccc4c13)CCC2=O	Cc1cc2c(c3ccc4ccccc4c13)CCC2=O	Cc1cc2c(c3ccc4ccccc4c13)CCC2=O	["Cc1cc2c(c3ccc4ccccc4c13)CCC2=O"]	valid	toximol_ames	toximol_ames/inputs.csv	19	1
25	ames	Nc1ccc2nc3ccccc3nc2c1	Nc1ccc2nc3ccccc3nc2c1	Nc1ccc2nc3ccccc3nc2c1	["Nc1ccc2nc3ccccc3nc2c1"]	valid	toximol_ames	toximol_ames/inputs.csv	25	1
37	ames	O=[N+]([O-])c1ccc2c3c(cccc13)-c1ccccc1-2	O=[N+]([O-])c1ccc2c3c(cccc13)-c1ccccc1-2	O=[N+]([O-])c1ccc2c3c(cccc13)-c1ccccc1-2	["O=[N+]([O-])c1ccc2c3c(cccc13)-c1ccccc1-2"]	valid	toximol_ames	toximol_ames/inputs.csv	37	1
40	ames	O=C(/N=c1\sn(C(=O)c2ccc([N+](=O)[O-])cc2)c2ccc([N+](=O)[O-])cc12)c1ccc([N+](=O)[O-])cc1	O=C(/N=c1\sn(C(=O)c2ccc([N+](=O)[O-])cc2)c2ccc([N+](=O)[O-])cc12)c1ccc([N+](=O)[O-])cc1	O=C(/N=c1\sn(C(=O)c2ccc([N+](=O)[O-])cc2)c2ccc([N+](=O)[O-])cc12)c1ccc([N+](=O)[O-])cc1	["O=C(/N=c1\\sn(C(=O)c2ccc([N+](=O)[O-])cc2)c2ccc([N+](=O)[O-])cc12)c1ccc([N+](=O)[O-])cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	40	1
55	ames	Nc1c2ccccc2nc2ccccc12	Nc1c2ccccc2nc2ccccc12	Nc1c2ccccc2nc2ccccc12	["Nc1c2ccccc2nc2ccccc12"]	valid	toximol_ames	toximol_ames/inputs.csv	55	1
58	ames	O=C1c2ccccc2C(=O)c2c1cc(O)c(CO)c2O	O=C1c2ccccc2C(=O)c2c1cc(O)c(CO)c2O	O=C1c2ccccc2C(=O)c2c1cc(O)c(CO)c2O	["O=C1c2ccccc2C(=O)c2c1cc(O)c(CO)c2O"]	valid	toximol_ames	toximol_ames/inputs.csv	58	1
59	ames	Cc1c2ccccc2c(C)c2ccccc12	Cc1c2ccccc2c(C)c2ccccc12	Cc1c2ccccc2c(C)c2ccccc12	["Cc1c2ccccc2c(C)c2ccccc12"]	valid	toximol_ames	toximol_ames/inputs.csv	59	1
3	ames	O=[N+]([O-])c1ccc([N+](=O)[O-])cc1	O=[N+]([O-])c1ccc([N+](=O)[O-])cc1	O=[N+]([O-])c1ccc([N+](=O)[O-])cc1	["O=[N+]([O-])c1ccc([N+](=O)[O-])cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	3	1
9	ames	Cc1c2ccccc2cc2c3c(ccc12)C(O)C(O)C=C3	Cc1c2ccccc2cc2c3c(ccc12)C(O)C(O)C=C3	Cc1c2ccccc2cc2c3c(ccc12)C(O)C(O)C=C3	["Cc1c2ccccc2cc2c3c(ccc12)C(O)C(O)C=C3"]	valid	toximol_ames	toximol_ames/inputs.csv	9	1
12	ames	Nc1ccc([N+](=O)[O-])cc1[N+](=O)[O-]	Nc1ccc([N+](=O)[O-])cc1[N+](=O)[O-]	Nc1ccc([N+](=O)[O-])cc1[N+](=O)[O-]	["Nc1ccc([N+](=O)[O-])cc1[N+](=O)[O-]"]	valid	toximol_ames	toximol_ames/inputs.csv	12	1
13	ames	O=[N+]([O-])c1ccc(/C=C/c2ccccc2)cc1	O=[N+]([O-])c1ccc(/C=C/c2ccccc2)cc1	O=[N+]([O-])c1ccc(/C=C/c2ccccc2)cc1	["O=[N+]([O-])c1ccc(/C=C/c2ccccc2)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	13	1
17	ames	O=C(c1ccccc1)C1OC1c1ccccc1	O=C(c1ccccc1)C1OC1c1ccccc1	O=C(c1ccccc1)C1OC1c1ccccc1	["O=C(c1ccccc1)C1OC1c1ccccc1"]	valid	toximol_ames	toximol_ames/inputs.csv	17	1
21	ames	NC(CCC(=O)NC(CSCCCl)C(=O)NCC(=O)O)C(=O)O	NC(CCC(=O)NC(CSCCCl)C(=O)NCC(=O)O)C(=O)O	NC(CCC(=O)NC(CSCCCl)C(=O)NCC(=O)O)C(=O)O	["NC(CCC(=O)NC(CSCCCl)C(=O)NCC(=O)O)C(=O)O"]	valid	toximol_ames	toximol_ames/inputs.csv	21	1
22	ames	O=[N+]([O-])c1ccc2c(c1)Cc1cc([N+](=O)[O-])ccc1-2	O=[N+]([O-])c1ccc2c(c1)Cc1cc([N+](=O)[O-])ccc1-2	O=[N+]([O-])c1ccc2c(c1)Cc1cc([N+](=O)[O-])ccc1-2	["O=[N+]([O-])c1ccc2c(c1)Cc1cc([N+](=O)[O-])ccc1-2"]	valid	toximol_ames	toximol_ames/inputs.csv	22	1
23	ames	CC(C)CC(=O)/N=c1\sn(C(=O)CC(C)C)c2ccc([N+](=O)[O-])cc12	CC(C)CC(=O)/N=c1\sn(C(=O)CC(C)C)c2ccc([N+](=O)[O-])cc12	CC(C)CC(=O)/N=c1\sn(C(=O)CC(C)C)c2ccc([N+](=O)[O-])cc12	["CC(C)CC(=O)/N=c1\\sn(C(=O)CC(C)C)c2ccc([N+](=O)[O-])cc12"]	valid	toximol_ames	toximol_ames/inputs.csv	23	1
24	ames	O=[N+]([O-])c1cccc2ccccc12	O=[N+]([O-])c1cccc2ccccc12	O=[N+]([O-])c1cccc2ccccc12	["O=[N+]([O-])c1cccc2ccccc12"]	valid	toximol_ames	toximol_ames/inputs.csv	24	1
28	ames	c1ccc2c(c1)-c1cc3ccc4ccccc4c3cc1C1NC21	c1ccc2c(c1)-c1cc3ccc4ccccc4c3cc1C1NC21	c1ccc2c(c1)-c1cc3ccc4ccccc4c3cc1C1NC21	["c1ccc2c(c1)-c1cc3ccc4ccccc4c3cc1C1NC21"]	valid	toximol_ames	toximol_ames/inputs.csv	28	1
31	ames	CCCCOCC1CO1	CCCCOCC1CO1	CCCCOCC1CO1	["CCCCOCC1CO1"]	valid	toximol_ames	toximol_ames/inputs.csv	31	1
33	ames	CN(C)CCNC(=O)c1cccc2cc3ccccc3nc12	CN(C)CCNC(=O)c1cccc2cc3ccccc3nc12	CN(C)CCNC(=O)c1cccc2cc3ccccc3nc12	["CN(C)CCNC(=O)c1cccc2cc3ccccc3nc12"]	valid	toximol_ames	toximol_ames/inputs.csv	33	1
38	ames	CCCCON(OC(C)=O)C(=O)c1ccc(-c2ccccc2)cc1	CCCCON(OC(C)=O)C(=O)c1ccc(-c2ccccc2)cc1	CCCCON(OC(C)=O)C(=O)c1ccc(-c2ccccc2)cc1	["CCCCON(OC(C)=O)C(=O)c1ccc(-c2ccccc2)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	38	1
41	ames	Nc1ccc(-c2ccc(N)cc2[N+](=O)[O-])c([N+](=O)[O-])c1	Nc1ccc(-c2ccc(N)cc2[N+](=O)[O-])c([N+](=O)[O-])c1	Nc1ccc(-c2ccc(N)cc2[N+](=O)[O-])c([N+](=O)[O-])c1	["Nc1ccc(-c2ccc(N)cc2[N+](=O)[O-])c([N+](=O)[O-])c1"]	valid	toximol_ames	toximol_ames/inputs.csv	41	1
43	ames	CC(=O)Nc1ccc2c(c1)Cc1ccccc1-2	CC(=O)Nc1ccc2c(c1)Cc1ccccc1-2	CC(=O)Nc1ccc2c(c1)Cc1ccccc1-2	["CC(=O)Nc1ccc2c(c1)Cc1ccccc1-2"]	valid	toximol_ames	toximol_ames/inputs.csv	43	1
45	ames	CC(=O)Nc1ccc(Oc2ccc(N(O)C(C)=O)cc2)cc1	CC(=O)Nc1ccc(Oc2ccc(N(O)C(C)=O)cc2)cc1	CC(=O)Nc1ccc(Oc2ccc(N(O)C(C)=O)cc2)cc1	["CC(=O)Nc1ccc(Oc2ccc(N(O)C(C)=O)cc2)cc1"]	valid	toximol_ames	toximol_ames/inputs.csv	45	1
47	ames	CCN(CCCl)CCCNc1c2ccc(Cl)cc2nc2ccc(OC)cc12	CCN(CCCl)CCCNc1c2ccc(Cl)cc2nc2ccc(OC)cc12	CCN(CCCl)CCCNc1c2ccc(Cl)cc2nc2ccc(OC)cc12	["CCN(CCCl)CCCNc1c2ccc(Cl)cc2nc2ccc(OC)cc12"]	valid	toximol_ames	toximol_ames/inputs.csv	47	1
51	ames	O=NN1CCN(N=O)CC1	O=NN1CCN(N=O)CC1	O=NN1CCN(N=O)CC1	["O=NN1CCN(N=O)CC1"]	valid	toximol_ames	toximol_ames/inputs.csv	51	1
56	ames	CC(=O)ON(C(C)=O)c1ccc2c(c1)Cc1ccccc1-2	CC(=O)ON(C(C)=O)c1ccc2c(c1)Cc1ccccc1-2	CC(=O)ON(C(C)=O)c1ccc2c(c1)Cc1ccccc1-2	["CC(=O)ON(C(C)=O)c1ccc2c(c1)Cc1ccccc1-2"]	valid	toximol_ames	toximol_ames/inputs.csv	56	1

ToxiMol structure-level detoxification

Toxic molecules to be repaired into non-toxic analogues that keep the core structure, one config per toxicity endpoint, with rendered molecule images.

Part of the Molecular Datasets collection.

Get started

from datasets import load_dataset
dataset = load_dataset("liuganghuggingface/moldata-toximol", "ames")

Where the data comes from

Original data: https://huggingface.co/datasets/HydroSophyTech/ToxiMol-benchmark
Original study: Breaking Bad Molecules: Are MLLMs Ready for Structure-Level Molecular Detoxification? (KDD 2026)
Built on: Therapeutics Data Commons

Tables

Config	Contents	Grade	Splits	Rows	Targets
`ames`	ToxiMol ames structure-level detoxification	ready	full	60	none
`carcinogens_lagunin`	ToxiMol carcinogens_lagunin structure-level detoxification	ready	full	60	none
`clintox`	ToxiMol clintox structure-level detoxification	ready	full	60	none
`dili`	ToxiMol dili structure-level detoxification	care	full	60	none
`herg`	ToxiMol herg structure-level detoxification	care	full	60	none
`herg_central`	ToxiMol herg_central structure-level detoxification	care	full	60	none
`herg_karim`	ToxiMol herg_karim structure-level detoxification	ready	full	60	none
`ld50_zhu`	ToxiMol ld50_zhu structure-level detoxification	ready	full	60	none
`skin_reaction`	ToxiMol skin_reaction structure-level detoxification	ready	full	60	none
`tox21`	ToxiMol tox21 structure-level detoxification	ready	full	60	none
`toxcast`	ToxiMol toxcast structure-level detoxification	ready	full	60	none

data/ contains rows with readable molecular structures; identical observations are collapsed. observations/ contains all source rows.

Source label columns are text, including values such as <10. SMILES_original retains the input representation; SMILES_canonical provides a canonical representation. structure_key groups molecular components, and structure_status records parsing outcomes.

Files and modalities

Modality	Where	Content
images	`images/ames/ames.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/carcinogens_lagunin/carcinogens_lagunin.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/clintox/clintox.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/dili/dili.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/herg/herg.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/herg_central/herg_central.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/herg_karim/herg_karim.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/ld50_zhu/ld50_zhu.parquet` (1 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/skin_reaction/skin_reaction.parquet` (1 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/tox21/tox21.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.
images	`images/toxcast/toxcast.parquet` (2 MB)	Original benchmark task parquet, unchanged, containing image bytes, molecule id, task and SMILES. Join to the input table on `id` and `smiles`; each image belongs to that same row. These are starting molecules, not experimentally validated repaired targets. Source-specific license notes remain on the dataset card.

Download an individual file:

from huggingface_hub import hf_hub_download
path = hf_hub_download("liuganghuggingface/moldata-toximol", "images/ames/ames.parquet", repo_type="dataset")

Data quality

Each config is graded on its own (Grade column above); metadata.json lists each config's quality tags.

Dataset notes

Starting compounds reuse TDC toxicity sources; this is a new repair benchmark, not new experimental toxicity evidence.
Generated-molecule safety is evaluated by predictive oracles, not prospective toxicology experiments.
The release provides starting molecules and optimization tasks.